Links
REBEL10 impressions
This is a collection of quotes from REBEL10 end-users taken from public forums like the
Computer Chess Club and our own
REBEL-BOARD.
Names of people are deliberately surpressed expect from those people we have asked their permission to do so.
Rebel9 says 2580 on the CD. What speed PC is assumed here? And is 2580 based on comp-human or comp-comp result? Appreciate any enlightenment here.
I played only 2 matches at 40/40 on 2 PII-400 with 256 MB RAM. Rebel 10 beat Fritz 5 20-18 and tied with Junior 5 10-10.
Rebel 10 feels stronger than Rebel 9, but I don't know to what extent this will show in computer-computer games.
Enrique
If Enrique's test results hold up with more games and different programs.(Rebel 10 beating Fritz 5 20-18 and Drawing Junior 5 10-10) Rebel 10 looks to be as strong as any other top program in the computer vs computer arena. Add to this the great results playing GM Anand, Rebel 10 could be hard to over look as the best program overall in both areas. We shall see.
Mark Young
When I see real evidence this is wrong I'll certainly change my mind. But the evidence has to be doing respectably vs GM players, not trashing some FM/IM players... There's a difference.. And the games have to be 40 moves in two hours or similar time controls...
Rebel 9 I guess is the best evidence I can think of that todays programs on todays computers are better then 2500 elo.(if thats the standard) At Ageon 97 Rebel 9 should have gone 4-0 Vs the IM's, but did score 3.5-.5. And did score 1-1 vs the GM's. No the games were not at 40/2 but it was slow enough to be called real tournament chess I think.
Then kick in Rebel 10's score of .5-1.5 playing GM Anand a Super GM at 40/2hr. And remember GM Anand has not allowed a draw from any IM player over the past few years. If the CCC post I read was correct in this fact.
I think this shows that todays programs are better then IM's. And by default the only thing better then an IM is a GM.
Ok the evidence is thin....but I took my best shot:)
Mark Young
at 40/120 rebel 10 whit 100 meg of memory on my amd k6 400 beat fritz 5 6 out of 10 game it did not do as well at game 60 or game 30 but at 40/120 it will be the best out it is hard to put the move in buy hand but i did it
i only played game whit fritz5 because it is number 1 but i belive it will also beat hairs6 to i set up one of my computer to play day and night rebel10 vs rebel9 at 40/120 rebel10 did not loose a game it did give rebel9 some drawes
ed this is a gm program i think you
ps if any body do not belive this program is a gm i will play over the phone or email if you like if you belive you can beat it on my amd k6 computer
email address is jcook166@erols.com
To those interested in Rebel 10:
My USCF high is 18-something.
I thought Genius 2 played "correctly", like one of my favorite players, Karpov. I always tried to out-technique it, since this is my chess ideal. I often reached simplified, technical, equal looking positions, but winning was almost impossible. I eventually scored 1 win, 5 draws, 4 losses at 40/2, on a 486/66.
Against Rebel 10, I NEVER reach equal, simplified positions. That's the "bad news" for me. The good news is: I sometimes get "chances". These chances usually demand energetic exploitation of some disorder that sometimes appears. In this respect, play against Rebel is very, very much like a real, scrappy game against a human. But, so far, it ALWAYS resolves the mess in its favor. This is a different brand of chess than my ideal, and there's no escape! Against Rebel 10, a lot more is required, than merely trying to hold my position together. Maybe some of this "lot more" can be learned. Novag Expert did this to me, years ago, and I DID eventually surpass it at 40/2.
Rebel 10 "never" makes "dumb computer moves". (So far. I haven't seen an ending yet!)
1) I can not make your decision, only give you personal impressions (no definite result about playing strength based on 500 games or so).
2) I have begun testing Rebel 10 as well since short and will soon perform a little 10 games welcome tournament here too. But that's not enough for final judgements.
3) What I like about Junior5's playing style: - it plays positional chess with a lot of pressure in many games and in all phases of the game. Even it's losing games are most times enterprising games. It often came up with really surprising moves and plans both in attack and defense. I have a faible for about this kind of inventive chess since I learned to play, no matter if humans or computers play it.
4) What I like about Rebel10's playing style: - Rebel10 from my view plays less blundering moves than maybe *any* other program I know. It often plays a calm and steady, but very effective style, often underestimated especially by humans. Looking at the games against humans I always have the phantasy that they so to say wondered how they got into trouble at some point in the game. Ed's programs have been especially successful in human games above blitz and fast levels for quite a while. Just look at the Aegon performances during the last years and have a look at the playing style of these games: Rebel10 is a further improvement in that direction as fast as I can see (not just for it's impressive performance against Anand).
5) Concerning Computer-computer playing strength: I am convinced that Junior5 and Rebel10 will both play for the No.1 in the SSDF list. And I guess they soon will both see a third dangerous opponent with Hiarcs7.
So I cannot make any decision for you. I recommend you get as many games as already available from both Rebel10 and Junior5 and decide according to your personal taste. I am sure that pure playing *strength* will not be the main argument with two programs so close and so strong. Look what you like most and have fun with it.
What is the best chess program against human?
I think that Rebel10
Rebel had always good results against humans
I think that Rebel10 Rebel had always good results against humans
I agree. Rebel 10 has to be the clear favorite here. Any program that can go toe to toe with GM Anand at all time controls has my vote :)
Mark Young
Having recieved a FREE copy of the full version of Rebel10 as a part of being one of the lucky few who correctly predicted the Anand/Rebel match, I wish to publicly thank Ed for keeping his honorable word.
I have my copy of Rebel10 and could not be happier! It is truly a great program that goes above and beyond what is currently available in chess playing software.
I think you would say that of any program you received for free!!
Not at all, but even if this was the truth what's wrong with that? I have been a Rebel fan for many years now. My first "serious" Schroder program was "Gideon Pro" back in the early 90's.
Ed Schroder's commitment to produce the strongest/best chess program in the world for the serious chess enthusiast, is exemplified in his latest creation of the Rebel series. This is not a secret, and there are more devoted people than myself who will agree that Rebel is as good or better than any other chess program currently offered.
I read in Chess Life that the score of that match for standard time controls of 40/2 was Vishy 1.5 Rebel 0.5 . The reason Rebel 10 won was because there were other games played at time controls that favored the computer...2 games at 5 m 5s, and 2 others at a time control of 15 minutes....and that the computer dominated those games by scores of 1.5 to 0.5.
IMO rebel10 dominated the 40/120 too. vishy found a fire-exit-door.
but he was overplayed by the program.
I think just about any player in the world besides Kasparov and a very few others would be happy going 1 draw 1 loss Vs Super GM Anand. Let alone a microcomputer chess program playing at 40/2hr. If you look at the games there is no question that Rebel 10 whent toe to toe playing GM Anand at 40/2hr.
Depends on what you mean. I think the impression of the match is colored by the success at short time controls. Two games with 1/2 point can hardly be considered going 'toe to toe' unless you assume that victory is unlikely. Toe to toe would be 1:1, would it not? To my primitive way of thinking, 1.5:0.5 is a loss for the side with 0.5. Of course, the average 'Joe on the street' would probably be pretty happy with 1/2 point in such situations. I think any GM would want to win. Even someone two hundres points lower. That's one reason why they do so well. A drive to win. Even if they would modestly state that they would be happy with a draw and a loss, I doubt if they would mean it.
*I* would be happy with a single draw, or even lasting 20 moves and losing. But that's another story.
I look at it a bit differently because I played over many of the games GM Anand played Vs. the other top chess programs past and present. He rolled them all into tiny meat balls. It did not matter the time control he still crushed them all. I don't think before Rebel 10, GM Anand had ever lost a game to a computer at any time control. Let alone lose a match or almost lose a game at 40/2hr as was the case in the first game of Rebel 10 Vs. GM Anand at 40/2. I don't think I over state the case when I say Rebel 10 whent toe to toe. IMO
Mark Young
8/5Bp1/4P3/6pP/1b1k1P2/5K2/8/8 w - - 0 1
The key is Kg4
Programs want to play fxg5, because this wins a pawn, prepares to get a passer,etc.
I haven't seen detailed analysis, but I assume that this draws, because black gets back to h8 somehow, and then he can sacrifice his bishop for the e6 pawn eventually, and white is left with the wrong rook pawn.
Does anybody figure this position out? If so, how did you do it? Does it require speculative knowledge that might lead to a sub-optimal result in other cases?
Rebel 8, Rebel 9, and Rebel 10 are the only programs I know of that can find the key move in this position. Rebel 9, and Rebel 10 being the only ones that can find Kg4 in well under 10 min. and stick with.
Here is what Rebel 10 thinks of the position.
Kg4 found at ply 16 score 2.77
Kg4 Be7 Kf5 g4 Kxg4 Ke4 Be8
At ply 17, score now 3.20 same line of play.
I am not an expert, but I have seen this EOC thing working and it's really nice.
First, you must make sure that EOC use is set to Active or Strong (as you have already done). Now, the bit that was not obvious from the documentation:
Click on the Data menu item. Click on Enter Database. Some database will be displayed. You want this database to be GAMEBASE. If it is NOT GAMEBASE, click on the upper left menu item (DB-FILES) and select "Load database". A file requestor pops up on the right side of the screen. You should be displaying the GAMES directory already. If not, navigate to the games directory. Then, select gamebase.dat.
Now, the GAMEBASE.DAT (300,000 game monster) should be displayed in your window. Press ESC to return to the main screen where the chessboard is displayed.
Click on new game and begin! I recommend pressing Shift-F2 to display a layout that will let you look at what is stored in the EOC. As you play a game, you'll hear about 5 seconds of disk access per move as the program sifts through the database looking for the position in the database. If it is there, after a short while of searching, the influence of the EOC can be seen as moves that are found in there are given more credence by the program. Naturally, blunders (or moves that look to Rebel 10 like blunders) may not be selected still. I saw an exchange sac by Karpov that Rebel would not play even with EOC set to Strong. Usually, though, it works very nicely. A very ingenious way to extend a program without just taking the worlds games and creating a book from them (though this is possible too with EOC=book setting). The use of Active and Strong is much more intelligent, IMHO.
Does anyway care to explain the decrepancy between the rebel10 bench mark test and the ssdf? On my pentium 120 with 16mb of ram rebel scored a benchmark of 2354, this is over 100 pts below the ssdf list which rates rebel on my processor to be around 2470. How accurate is this benchmark test? It appears to me that the test is kind of persimmistic regarding the ratings of programs. Perhaps there is a secret robert hyatt algorithm written within the program!!!!! No offense bob just kidding!!
Wait what are you talking about 2354 rated players like Anand :), trash Rebel 10 all the time, wait i've lost my mind if i thout for a second rebel was only 2354. Rebel10 is a Grandmaster.
I think Rebel 10 witll be the Dark horse of the Tournament. I think it may hold some surprises for all to see.
Rebel I believe is a stayer, and will sprint ahead in the final strait GO REBEL !!!!!!!!
So here's the replacement of game3 of the Rebel welcome tournament, after there were doubtful moves in the autoplayer score of the original game 3. Once more Rebel game won by excellent positional skill.
It is games like this that shaped my response to the CCC opinion Poll on GM's recognizing the play of a computer. In that poll I went with they would after many games, not just a few. Rebel's pieces all landed on nice squares and the win seemed effortless. As you say a very nice positional win. The placement of the Rooks, the well placed Knight and the control of the centre. Not sure how one could label white's play as computer-like in this example.
I'm also worried about recent developments about the commercial future of chess programmers and resellers. But I think that a true crisis would only be on the horizon if Ed & Co. did run out of new ideas that justified spending money on ever new versions of their products (I think that even $100 is not too much to ask for a program of Rebel's virtues, for $60 it's a steal, that's very close to the $50 you have to pay e.g. for ChessMaster or ChessSystemTal here in Germany). Fortunately, the end of this road of convincing new features is not in sight by far.
Some examples:
Rebel 10 has several HUGE improvements over previous versions and also some "industry firsts":
1) EOC - chess tree used not only for reference (in a particularly attractive way, including reference to 60 "all time best" GMs who played certain variants) but also in addition to hand crafted opening book as a bonus & penalty guide for the engine to chose the right moves from GM theory.
2) Anti-GM option, user selectable degree of "Anti-GM"-ness, interesting idea, Anand-proven, works also in many other types of !! GM positions from my databases.
3) Screen layout presets via F1-F5 to manage information overflow - similar to CST, sounds simple, works great.
4) Automated test suite processing with many new parameters (extended EPD).
5) about 20 chess annotation symbols (sadly missing from previous Rebel versions).
6) several times faster analysis in analysis mode by using specialized algorithms - good for overnight analysis and test suites
7) Finally - Resizable chess board (also at 125% of the size you had in R9), resizable, moveable, configurable information windows (focus on the important things, forget the rest), together with 3) a real usability improvement.
8) Annotation on-the-fly during the game if you make a mistake (evaluation swing), you see an asterisk * beneath your questionable move and can jump back immediately after finishing (losing) the game.
9) Vastly extended database search (on dates, free index, ...)
10) Fischer clock increment (important at Blitz !!!), user definable 40/x minutes time control
etc. etc. etc.
And there are even some things left for Rebel 11:
a) variants
b) Windows NT compatibility
c) autoplayer support
d) endgame tablebases
e) chess coach like in Genius, Fritz, ... or preferrably even something better
f) logfile like in Crafty
g) ICC support
So I don't worry about innovation for the next couple of years ...
Everything I wrote about Rebel in my review on http://207.86.247.69/wccr/software/Rebel9/mb.html is still true.
Please note that Ed has incorporated many featured I missed in Rebel 9 in his latest release and also take a look at the numerous other improvements - small and big - he introduced with Rebel 10.
Consider also the HUGE amount of chess material, utilities etc. on the R10 CD and give him credit for love, passion, care and money that he and his team put into making R10 (and the Anand match, BTW) a reality for their customers.
Don't forget that Ed gives technical support for his products for free here on CCC and via email, that he shares his ideas and even offered his EOC builder utility (which he felt was not good enough for a public release by his standards) to any interested parties.
Now Ed made a decision to cut prices by 50%. Who will suffer most? Of course Ed himself if he doesn't sell at least twice as many copies of Rebel, which I certainly hope he will be able to do.
You simply cannot discount the fact that Ed offers Decade 2.0 for free and will also give you a free upgrade to the preliminary Windows version when it's ready. Don't put up a big show here that you didn't get enough value. I simply don't know any other chess program that will give you as many additional goodies (DBUTIL 2.5 etc.) as Rebel does, no matter if at $60 or $120.
Still, I hope Ed will find a way to give back something else than Rebel 1 to his most loyal customers who bought Rebel at full price. But even if he doesn't offer you more than what you already have - it makes no sense to get 0% of the fun at 100% of the price just because you're that much angry at having payed the same amount of money for Rebel like you probably did for Genius, Shredder, Nimzo, Hiarcs, M-Chess, Fritz, ...
If you want to recover your loss, why not play twice as often as you would have normaly done against Rebel, thus getting 200% value our of your purchase - hey, no ill will intended, I'm not mocking you, but don't spoil yourself the fun of owning one of the most enjoyable chess programs.
Famous last words ... I hope that I will be able to finish my R10 review within the next days, there I will also mention some things I don't like about R10, but with the new price tag I will probably elevate it from "BUY" to "STRONG BUY" (read: must have).
I should mention that there are other small differences between the 2 programs, but it is obvious that Junior 5 and Rebel 10 are the 2 top programs both in engine positional strength and most and best features. Congratulations to Amir and Shay and ChessBase; Ed and Rob of Scroder BV for 2 fantastic chess programs.
Hi CCC
I would like to share this exciting game with the group. In my view It is the most astonishing chess game between two computers I have ever seen!!. Rebel 6.0 although it is a now old program shows it still has sting. At move 8.. chessmaster goes out of book with d6 which I believe is a theoretical novelty. It then follows this idea up with the astonishing 10..Bh3!!! the most incredible move I have ever witness from a computer!. Also I discovered this position is a very good illustration of Rebel10 Anti Grandmaster Feature, Rebel9 does not find this move after ten minutes, yet rebel 10 on my cyrix233 finds the move in 30sec! I think that the position itself would be a classical example of an anti grandmaster move!! because the complications are so incredible, I doubt many humans could correctly navigate.
As I do not know the reasons why Al judged the playing strength of both program in this way, I propose he explains it himself, before we speculate over that.
So Korner, please explain to avoid msiunderstanding.
Up to now my impression (still based on Mcp7.1) was that while Mchess is the stronger blitzer against humans (with its powerful speculative attacking chess bringing humans to despair, Rebel has been the best program against humans on tournament level since years, playing the more mature positional chess at these time controls. Maybe this changed with Mchess8, but on which results over which time could such a judgement possibly be based?
I am a strong Class "A" player and I own Rebel 9. It is so strong that I only use it for analyzing my over-the-board games. In many cases it finds resources that neither myself or my opponent noticed during the game. I would agree that the Rebel 9 must be at GM strength or very close to GM strengh.
Hello all out there!
Sorry for the delay in reporting from our tournament, but I had some problems in getting access to the internet after upgrading my computer. Rebel 10 won the tournmant without loosing a single game!
Hiarcs played well, but was perhaps a little bit handicaped, because the hardware of the Hiarcs 7 "Bauer", K6-333 was out of order and crushed several times. But step for step We played a seven-round-swiss system tournament with three hours each side for the wohle game.
Every operator was allowed to change level or the internal clock three times a game to prevent time trouble. Other influences during a game were not allowed. We tried to let start each program in his newest (strongest?!) version and let start each one TWO times! So we got 14 games of each program, which gave a good picture of their abilities.
We made one exception for Nimzo 3,5, because the operator was making many additional programming by "Che" and it was promised to him in our last tournament, that he could use Nimzo 3.5 in the next one. In the final ranking you find names for differing the computers in the tournament, but the games are commentated with the hardware specifications.
Here you find the final result
FINAL RANKING after 7 rounds Cross-table (using entry numbers of players) Points Buchholz 1. 5 Rebel10 Bauer 5.5 25.5 2. 6 M-Chess Pro8 Fuhrwer 5.0 27.5 3. 1 Fritz5 Gnes 5.0 26.0 4. 8 Virtual Chess II Her 4.5 29.0 5. 17 Rebel10 Friedrich 4.5 28.5 6. 12 Fritz5 Poetsch 4.5 28.0 7. 2 Nimzo99 K„stner 4.5 22.5 8. 16 Hiarcs7 Mitsd”rfer 4.5 19.5 9. 9 Chessmaster 6000 Gru 4.0 30.0 10. 18 Zarkov4.3 Gnes 4.0 27.5 11. 24 Genius5 P60 4.0 26.5 12. 3 Junior5 Benstein 4.0 26.0 13. 21 Shredder3 Herklotz 4.0 25.5 14. 4 Hiarcs7 Bauer 4.0 25.5 15. 29 Kallisto II Serfling 4.0 22.5 16. 10 Nimzo3.5 Kästner 3.5 26.5 17. 26 Chess System Tal Fuh 3.5 23.0 18. 23 Chessmaster 6000 Rie 3.5 20.0 19. 28 Zarkov4.3 Keuter 3.0 28.5 20. 15 Junior5 Feuerbaum 3.0 23.5 21. 13 Chess System Tal Bau 3.0 21.5 22. 27 Gandalf3 K„stner 3.0 20.5 23. 14 Gandalf3 Didzuhn 2.5 28.5 24. 7 Shredder3 Bauermeist 2.5 26.0 25. 11 Genius5 P468 2.5 21.0 26. 22 Virtual Chess II Rie 2.5 19.0 27. 19 Kallisto II Cyrix 2.0 23.0 28. 20 M-Chess Pro8 Tauber 2.0Additionally we calculated the "team ranking", which was also won by Rebel 10 (nobody wonders!) in front of Fritz 5.32 and Hiarcs 7!
All participants were in agreement, that Rebel 10 played the best chess and won well-earned! We were a little bit disappointed by the play of Nimzo, Shredder and Junior, who were not playing as excepted... Hiarcs, Fritz and some others played well, but in this strong field nearly every one could win...
Positive surprises were Zarkov and the new Chessmaster!
Here is this ranking
1. Rebel 10 10/14 2. Fritz 5.32 9,5 3. Hiarcs 7 8,5 4. Nimzo 3.5 / 99 8 5. Chessmaster 7,5 6. Zarkov 4.3 7 7. M-Chess Pro 8 7 8. Virtual Chess 7 9. Junior 5 7 10. Shredder 3 6,5 11. Chess System Tal 6,5 12. Genius 5 6,5 13. Kallisto II 6 14. Gandalf 3 5,5 15. Tasc R30 Vers.2,5 2,5More you will find next days in the CCC or I will answer you, when you have more question.
Karsten Bauermeister
caratan@uni-muenster.de